ARTICLE IN PRESS
Planetary and Space Science 52 (2004) 1169–1175
www.elsevier.com/locate/pss
First observations of the Na exosphere of Mercury with the highresolution spectrograph of the 3.5M Telescopio Nazionale Galileo
Cesare Barbieria,, Stefano Verania, Gabriele Cremoneseb, Ann Spraguec,
Michael Mendillod, Rosario Cosentinoe, Donald Huntenc
a
Department of Astronomy, University of Padova, I-35122 Padova, Italy
b
Astronomical Observatory of Padova, INAF, Italy
c
Lunar and Planetary Laboratory, University of Arizona, Tucson, AZ 85721, USA
d
Center for Space Physics, Boston University, MA 02215, USA
e
Telescopio Nazionale Galileo, INAF, Italy
Received 16 September 2003; received in revised form 14 June 2004; accepted 6 July 2004
Abstract
Analysis of three spectra of the exosphere of Mercury in the Na-D lines are presented. Spectra were secured with the highresolution spectrograph (SARG) of the 3.5 M Telescopio Nazionale Galileo (TNG, located on the Roque de los Muchachos,
Canaries) on the evenings of August 23 and 24, 2002. Spectra have resolution R ¼ 115 000; the slit length was 26.7’’. The Na column
abundances range from 4.3 to 19:7 1010 atoms=cm2 with the highest abundances being close to the illuminated limb. Our
observations at true anomaly angles (TAA) from 1711to 1741show the traces of the emission lines to be strongly peaked at the
illuminated limb, supportive of recent modeling that shows thermal desorption to be a strong factor in determining the distribution
of Na about the planet.
r 2004 Elsevier Ltd. All rights reserved.
Keywords: Mercury; Planetary exospheres; Na emission atmospheres; High-resolution planetary spectroscopy
1. Introduction
The existence of the atmosphere around Mercury was
discovered by the Mariner 10 spacecraft, which revealed
UV emissions of three atomic elements: H, He and O
(Broadfoot et al., 1976). Three other elements (Na, K,
and Ca) were later discovered with ground-based
observations (Potter and Morgan, 1985; Potter and
Morgan, 1986; Bida et al., 2000, respectively). Due to
the low density (of about n ¼ 105 atoms=cm3 ; P ¼ 1012
bar, on the dayside), the atmosphere is collisionless, i.e.,
the mean-free path of the atoms is longer than the value
of the scale height H of the atmosphere (cf. Chamberlain and Hunten, 1987). Therefore, the whole atmoCorresponding author. Fax: +39-498278245.
E-mail address: [email protected] (C. Barbieri).
0032-0633/$ - see front matter r 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.pss.2004.07.013
sphere is comparable with an exosphere having the
exobase coincident with the planet’s surface.
The lifetime for the species in the exosphere is ruled
by the interaction with the interplanetary medium.
Photoionization is the fastest loss mechanism of the
neutral atmosphere: the ionized atoms are driven
away from the planet by the solar wind, or aimed back
to the surface along the lines of Mercury’s magnetic
field. To maintain the exosphere, the lost atoms must
be replaced by some source mechanisms. Processes of
endogenic and exogenic origin are supposed to act in
repopulating the exosphere. The expected sources of
Na include thermal, photon, and electron-stimulated
desorption, ion and chemical sputtering, vaporization
of regolith and of projectiles following micrometeoritic impacts (cf. McGrath et al., 1986, Hunten
and Sprague, 1997).
ARTICLE IN PRESS
1170
C. Barbieri et al. / Planetary and Space Science 52 (2004) 1169–1175
As gas-phase collisions are negligible, each constituent of Mercury’s atmosphere forms an exosphere having
different parameters (T; H; n) with respect to the other
species. This means that it is possible to follow the
evolution of a particular exospheric component without
taking into account the others. Exospheric atoms are
liberated from the surface with a supra-thermal energy
(i.e. with T gas 4T surf ), which depends on the nature of
their source mechanism. They can gravitationally escape
from the planet (escape velocity ¼ 4:3 km=s), or go into
a ballistic orbit which will re-encounter the planet’s
surface. After a bounce, the atoms can stick to the
surface, or bounce again into another ballistic orbit with
a different energy, closer to the temperature of the
surface, until they are photoionized or readsorbed by
the surface (cf. Hunten et al., 1988). The number and the
path of the bounces is determined by the nature of the
gas–surface interactions, by the lifetime of the neutral
atom in the exosphere, and of course by the gravity and
radiation pressure. As the atoms can perform several
bounces, it is expected that a thermal component is
formed, even if the source was supra-thermal, like on the
Moon. A thermal component is also liberated from the
surface when temperatures are high and the lifetime on
the surface is short. At Mercury the thermal component
should be the dominant one, because of the higher
temperature and greater gravity (cf. Hunten and
Sprague, 1997). Indeed, a measurement of the temperature of the Na component via high-resolution spectroscopy was performed by Killen et al. (1999a). They
found the presence of two components, a supra-thermal
one (1500 K) and another one (600 K) much closer to the
surface’s temperature.
Following the discoveries of Na and K at Mercury,
many ground-based observations have provided the
following information regarding their distribution and
time variation:
rapid time variations in the intensity of the exosphere,
not correlated with changes in solar activity (e.g.
Sprague et al., 1997, Potter et al., 1999),
enhancements of sodium emission over known
geological features on Mercury’s surface (Caloris
Basin, the Kuiper Murasaki crater complex), and
over regions which show radar-bright spots, associated with regions of fresh excavation features on the
side of Mercury not imaged by Mariner 10 (Sprague
et al., 1998),
a sodium comet-like tail streaming in the antisunward direction with variable abundance and
extent (Potter et al., 2002).
Source mechanisms for Na have long been debated.
Sprague et al. (1997) found a strong diurnal variation in
the emission of Na and K, with a large enhancement in
the morning. They conclude that a recycling mechanism
is at work, such that some photoionized atoms are
driven back to the surface along the lines of the
magnetosphere, then neutralized and adsorbed to the
surface in the night hemisphere, then released back in
the exosphere with the morning illumination. A similar
morning-side enhancement has also been observed in the
data of Potter (2003, Pers. comm.). Enhancements of K
over Caloris Basin and Na over the radar-bright spots
were interpreted by Sprague et al. (1998) to be further
evidence that diffusion and thermal desorption from
surface materials is a primary source. On the other
hand, time variations in the abundance of Na and
spottiness of the emissions observed in imaging were
interpreted to be evidence for charged particle sputtering
by Potter and Morgan (1990). A later data set showing
changes in Na emissions from the northern to southern
latitudes was interpreted by Potter et al. (1999) as
evidence for the effects of space weather on the
distribution and time variation of Na.
Several models of Mercury’s Na exosphere have
explored the distribution of Na around the planet under
the influence of selected physical processes known to be
working in the vicinity of Mercury. Some of these are
charged particle sputtering (Killen et al., 1999b),
radiation acceleration (Ip, 1990; Smyth and Marconi,
1995), ion implant followed by diffusion and thermal
evaporation (Sprague, 1992). A more holistic approach
including charged particle sputtering, photo desorption,
and thermal desorption using a constant source of
micrometeoritic bombardment generated Na has recently been published (Leblanc and Johnson, 2003). The
Leblanc and Johnson approach resulted in unique
distributions of Na about the planet as a function of
true anomaly angle (TAA). The model predicted the Na
tail at the TAA observed by Potter et al. (2002).
To provide more data to help understand the
phenomena described above, we have undertaken
observations with the SARG/TNG. The advantages of
using the TNG in this context come from the good
collecting aperture, from the outstanding image quality
of the site and of the telescope, and from the excellent
performances of the SARG in terms of efficiency and
resolution. This paper presents the results obtained in
the first, largely exploratory, attempt to reach this goal.
We judge the results promising enough to plan a
program of observations for the next several years.
Not only do we hope to clarify the relative role of source
mechanisms but we also hope to provide a useful data
bank for the forthcoming space missions MESSENGER
and BepiColombo.
2. Observations and data reduction
The SARG has been equipped with a Na filter
specifically for this purpose of studying diffuse Na in
ARTICLE IN PRESS
C. Barbieri et al. / Planetary and Space Science 52 (2004) 1169–1175
Solar System objects, because it allows to keep a long
slit (26.7 arcsec) on the sky by removing order overlapping. The main characteristics of the spectrograph
are given in Table 1.
For further information on SARG, see http://
www.pd.astro.it/sarg/. The observations discussed here
were carried out on two evenings: 23 and 24 Aug 2002
between 19:00 and 20:00 UT, i.e. during daytime or with
the Sun just below the horizon, the minimum elevation
of the TNG being 13.51. To save read-out time, the slit
was 2-binned in the spatial direction making the
effective spatial pixel of 0.64 arcsec. The evening of the
23rd was the best in terms of sky transparency and
seeing quality.
Mercury and Venus were both near eastern elongation, permitting us to obtain, sequentially and essentially
at the same air mass, spectra of Mercury, and Venus
across the terminator as null reference for Na emission,
and of Spica for flux and point spread function (PSF)
calibration. Fig. 1 shows an example of each spectrum,
and Fig. 2 the PSF of Spica.
Fig. 3 shows on the same scale the planet and the slit,
here in PA = 901. The Sun is in the upper right corner.
Relevant geometric and illumination parameters for our
observations are given in Table 2.
Reference spectra (flat fields, bias, dark current,
Th–Ar reference lamp) were taken both before and
after Mercury’s observations and used with the Interactive Reduction and Analysis Facility (IRAF) to
prepare the images for analysis using the standard
techniques of spectral image reduction. Particular care
was adopted in subtracting the diffuse light and the
1171
Fig. 2. Profile of Spica during the night of the 23rd. The FWHM was
1.0’’.
Fig. 3. Mercury and slit (to scale, here shown parallel to PA=901).
North is up, East to the left.
Table 2
Geometric and illumination parameters
Table 1
Instrumentation parameters
Spectrograph resolution
Slit length and width
Pixel dimension and scale
CCD dimension
23 Aug. 2002 24 Aug. 2002
115 000
26:7 0:40 arcsec
0.022 A, 0.32 arcsec
2K 4K pixels
Fig. 1. Spectra of Mercury (top), Venus across the terminator
( and D1 (5896 A)
(
(middle), and Spica. On Mercury, the D2 (5890 A)
lines are clearly visible in emission inside the broad Na Fraunhofer
absorptions seen in the reflected continuum from Mercury surface.
UT
Heliocentric distance (AU)
Phase angle
Exposure time
Diameter (arcsec)
Heliocentric Doppler shift (m)
Sub-Earth longitude, latitude (degrees)
True anomaly (degrees)
19:34
0.464866
71.0
30 s
6.3
34.4
148.6, 6.5
171
19:47, 19:49
0.465744
72.6
30, 60 s
6.4
24.8
153.5, 6.6
174
Mercury continuum, in order to obtain the two emission
lines from the exosphere. Fig. 4 shows sample spectra
from Mercury showing the Na D2 and D1 emission lines
on the short-wavelength side of the solar Fraunhofer
reflected from Mercury’s surface. Also shown are a
reflected solar spectrum and an enlarged portion of the
Mercury spectrum showing the D2 emission line and the
estimated continuum which was removed from the data
in order to obtain the uncontaminated emission lines.
ARTICLE IN PRESS
C. Barbieri et al. / Planetary and Space Science 52 (2004) 1169–1175
1172
1.5x104
1.2x104
1.0x104
8.0x103
counts
counts
4
1.0x10
5.0x103
6.0x103
4.0x10
3
2.0x103
0
5884
5886
5888
5890
5892
5894
5896
0
5884 5886 5888 5890 5892 5894 5896 5898
5898
wavelengh (A)
wavelengh (A)
Mercury Exosphere 23 Aug 2002
600
2.5
400
1.5
counts
Intensity (MR)
2.0
1.0
200
0.5
0.0
− 0.5
5886
5888
5890
5892
5894
5896
5898
5900
wavelengh (A)
0
5889.3
5889.4
5889.5
5889.6
wavelengh (A)
Fig. 4. Upper left: Mercury þ sky þ diffuse light (upper line) and sky + diffuse light only (bottom line). Upper right: after sky and diffuse light
subtraction, the solar reflected continuum and the exospheric D1 and D2 emission are left. Bottom panel: left, after solar continuum removal; right,
continuum estimate under D2.
3. Data analysis and results
The position angle (PA) of the slit was at a different
value for each of the three spectra presented here,
namely PA 901, 1521, and 1181, respectively. An
equatorially aligned slit corresponds to PA 1181. Fig. 5
shows the traces of instrumental data numbers (DNs)
along the slit for the continuum, along with a seeing
smeared model of the trace across the slit, and for the
D2 and D1 lines. The purpose of the seeing smeared
model trace is two-fold. First, it provides a measure of
the atmospheric ‘‘seeing’’ during the exposure because it
can be compared directly to the continuum trace along
the slit. Second, it provides a Hapke rough-reflectance
surface model (Hapke, 1986) composed of the rough
reflectance (Rr) at each location in the smeared surface
grid. The grid size can be chosen to represent the
parameters of the spectrograph (in this case a spatial
pixel of 0.32 arcsec binned 2; namely 0.64 arcsec). The
Hapke Rr can then be used to absolutely calibrate the
data using an average of the maximum values from the
continuum at a wavelength close to the sodium D lines.
We have used the method of Sprague et al. (1997).
For estimating the seeing, we generated several
models of each spectrum, then extracted the Rr
corresponding to the location of the slit at the time of
observation. By plotting the model trace along the slit
on the same graph as the actual data, it was possible to
get a good estimate of the seeing at the time of the
exposure, as shown by the dotted lines in Fig. 5.
We then partitioned the DNs of the D2 and D1
emission along the slit into sectors that are reasonable,
given the size of the planet and the actual seeing smear
of the image. In these three slit spectra we chose three
sectors along the slit for individual conversion to
column abundance. With the aid of the Hapke model
of the planet, for each exposure we determined the
approximate distance of the slit sector in units of
fraction of a radius from the center of the planet. From
this we also estimated the approximate longitude and
latitude of the central portion of the slit sector under
analysis. The radiative transfer program accounts for
illumination and emission geometry and seeing smear
(Hunten and Wallace, 1993; Sprague et al., 1997). For
spectral images of excellent seeing and higher spatial
resolution, it will be necessary to account for changing
surface albedo beneath the slit in the absolute calibration of the slit sectors before conversion of emission
DNs to column abundances (Domingue et al., 1997).
Table 3 shows g-factors, Rr, and calibration factors
for the three spectra presented in this paper. We use the
standard definitions of all the parameters, common in
ARTICLE IN PRESS
C. Barbieri et al. / Planetary and Space Science 52 (2004) 1169–1175
23 August 2002, 19:49 UT
Data Numbers
1.2.10
1.104
1.2.104
8.103
6.103
4.103
cont
scmod
d2
d1
2.103
0
-10
4
1.10
24 August 2002, 19:49 UT
24 August 2002, 19:47 UT
1.4.104
seeing sigma 1.3 arcsec
PA 90 deg
30 sec exp time
Data Numbers
4
1.104
seeing sigma 2.1 arcsec
PA 152 deg
30 sec exp time
8.103
6.103
8.103
Data Numbers
1.4.104
cont
scmod
d2
d1
4.103
10
0
-10
seeing sigma 2.2 arcsec
PA 118 deg
30 sec exp time
6.103
4.103
cont
scmod
d2
d1
2.103
2.103
-5
0
5
Arcsec from disk center
1173
0
-5
0
5
Arcsec from disk center
10
-10
-5
0
5
Arcsec from disk center
10
Fig. 5. DN along the slit are shown for all three spectral images presented in this paper. DNs for the continuum are obtained from 1 pixel in the
wavelength dimension (0.022 arcsec) and 2 pixels (0.64 arcsec) in the spatial direction. For the D2 and D1 emission line traces, DNs are from 7 pixels
(0.154 arcsec) in the spectral and 2 pixels (0.64 arcsec) in the spatial dimension. Also shown (dotted line, scmod) is the trace across the seeing smeared
Hapke model for each spectrum.Left: Model indicates good atmospheric seeing ðs ¼ 1:3 arcsecÞ for day 1. The slit was oriented 281with respect to the
equator, with the slit being North of the equator coming across the terminator, crossing the equator and going slightly South on the sunlit limb. The
emission is offset from the continuum toward the illuminated limb of the planet. Center: The slit was oriented 621with respect to the equator with the
left-hand side being South East in the sky, or coming from the South across the right side of the planet across the terminator, across the equator and
going North toward the bright limb. The continuum is fitted reasonably well with s ¼ 2:1 arcsec: The contrast of the smooth continuum to the bumpy
D2 and D1 emission indicated patchy Na emission. This is discussed in the text.Right: The slit was oriented along the planet’s equator, the left-hand
side being the night side of the planet. While the emission lines show brighter emission over the bright limb than the terminator, the seeing smear of
the continuum ðs ¼ 2:2 arcsecÞ does not exhibit the same behavior. This is discussed in the text.
Table 3
Absolute calibration parameters
23 Aug. 19:49
24 Aug. 19:47
24 Aug. 19:49
Exp. time (s)
Doppler shift (m)
g1 ðs1 Þ
g2 ðs1 Þ
Hapke Rr
Cal. factor (kR s DN1 )
30
30
60
34.4
24.8
24.8
1.7
1.5
1.5
2.7
2.5
2.5
0.0101
0.0078
0.0074
1.6
1.1
3.1
airglow work detailed in Hunten et al. (1956) and
Chamberlain and Hunten (1987). From the observed
D2/D1 intensity ratios, it is evident that the Na
exosphere is optically thick.
Table 4 gives the approximate locations of the slit
sectors, the absolutely calibrated brightness and the
vertical column abundance for each emission line.
equatorial values is observed. There is no statistically
significant difference between the mid-a.m. and the midday sectors, while the early a.m. sector exhibits slightly
lower abundances in accordance with the results of
Sprague et al. (1997). These results are also plotted in
Fig. 6.
5. Interpretations and conclusions
4. Discussion
The column abundances found in this new observing
campaign are very consistent with those of previous
observers (cf. Potter and Morgan, 1986; Sprague et al.,
1997). The mid-afternoon and late-afternoon portions
of the illuminated disk were not available during these
observations, but we explored the possibility of a
diurnal effect in the data by calculating the densities in
the early a.m., mid-a.m. and mid-day sectors (see Table
4). As can be seen in Table 4, the range of locations and
the samplings of data from North to South is scant, but
we plot the column abundances as a function of fraction
of radius from disk center to see if any strong North–South asymmetry is exhibited. A large scatter of
We have demonstrated that we are well equipped to
make high-quality observations using the TNG of
Mercury’s Na exosphere. Our results of these first
observations give column abundances and distribution
of Na consistent with those of previous observers.
Although our abundances show considerable scatter as
they are obtained from many locations on the disk and
the number of points is small, the column abundances
and the distribution of the Na about the planet are
consistent with the predictive modeling of Leblanc and
Johnson (2003), who found that column abundances are
highest in a band around the sunlit limb at TAA ¼ 1811
(see Plate 3, pag. 272). The true anomaly of our
observations varies from 1711to 1741. The emission
ARTICLE IN PRESS
C. Barbieri et al. / Planetary and Space Science 52 (2004) 1169–1175
1174
Table 4
Abundances
Location
X (E–W)
Brightness (kR)
Y (N–S)
lat (deg)
Column abundance (atoms
cm2 units of 1010 )
long (deg)
(fraction of radius from disk center)
Na Column Abundance
(units of 1010 atoms cm-2)
23 Aug. 19:49
early a.m.
early a.m.
mid-a.m.
mid-day
mid-day
24 Aug. 19:47
early a.m.
early a.m.
mid-a.m.
mid-a.m.
mid-day
24 Aug. 19:49
early a.m.
mid-a.m.
mid-a.m.
mid-day
0.29
0.00
0.30
0.76
0.90
0.16
0.00
0.21
0.41
0.44
12
6
2
18
25
160
150
135
105
75
D2
86.4
179.3
342.3
489.0
498.8
D1
65.2
102.7
200.5
277.1
285.2
D2
5.82
8.81
13.2
15.5
18.9
D1
5.69
6.86
10.6
10.3
10.6
0.25
0.00
0.30
0.60
0.72
0.25
0.0
0.2
0.4
0.65
10
6
15
30
40
160
150
140
125
85
148.2
242.8
285
322.6
311.2
114.0
185.8
201.8
225.7
231.4
9.63
13.3
12.1
12.7
14.4
9.76
13.9
12.0
11.9
12.8
0.30
0.10
0.60
0.95
0.00
0.00
0.00
0.00
6
6
3
0
160
145
110
85
180.7
289.2
365.8
381.3
144.8
237.5
316.2
325.5
14.3
17.4
18.9
19.6
14.0
16.4
17.2
18.8
20
References
15
Bida, T.A., Killen, R.M., Morgan, T.H., 2000. Discovery of calcium in
the Mercury’s exosphere. Nature 404, 159–161.
Broadfoot, A.L., Shemansky, D.E., Kumar, S., 1976. Mariner 10—
Mercury atmosphere. Geophys. Res. Lett. 3, 577–580.
Chamberlain, J.W., Hunten, D.M., 1987. Theory of planetary atmospheres: an introduction to their physics and chemistry. Int.
Geophysics Ser. 36.
Domingue, D.L., Sprague, A.L., Hunten, D.M., 1997. Dependence of
mercurian atmospheric column abundance estimations on surfacereflectance modeling. Icarus 128, 75–82.
Hapke, B., 1986. Bidirectional reflectance spectroscopy: 4. The
extinction coefficient and the opposition effect. Icarus 67, 264–280.
Hunten, D.M., Sprague, A.L., 1997. Origin and character of the lunar
and mercurian atmospheres. Adv. Space Res. 1551–1560.
Hunten, D.M., Wallace, L.V., 1993. Resonance scattering by
mercurian sodium. Astrophys. J. 417, 757–761.
Hunten, D.M., Roach, F.E., Chamberlain, J.W., 1956. A photometric
unit for the airglow and aurora. J. Atmos. Terr. Phys. 8, 345–346.
Hunten, D.M., Morgan, T.H., Shemansky, D.E., 1988. The Mercury
Atmosphere, Mercury. University of Arizona Press, pp. 562–612.
Ip, W.H., 1990. On solar radiation-driven surface transport of sodium
atoms at Mercury. Astrophys. J. 356, 675–681.
Killen, R.M., Potter, A.E., Fitzsimmons, A., Morgan, T.H., 1999a.
Sodium D2 line profiles: clues to the temperature structure of
Mercury’s exosphere. Planet. Space Sci. 47, 1449–1458.
Killen, R.M., Potter, A.E., Morgan, T.H., 1999b. Spatial distribution
of sodium vapor in the atmosphere of Mercury. Icarus 85,
145–167.
Leblanc, F., Johnson, R.E., 2003. Mercury’s sodium exosphere. Icarus
164 (2), 261–281.
McGrath, M., Johnson, R.E., Lanzerotti, L.J., 1986. Sputtering of
sodium on the planet Mercury. Nature 323, 694–696.
Potter, A.E., Morgan, T.H., 1985. Discovery of sodium in the
atmosphere of Mercury. Science 229, 651–653.
Potter, A.E., Morgan, T.H., 1986. Potassium in the atmosphere of
Mercury. Icarus 67, 336–340.
10
5
0
early a.m.
mid a.m.
mid - day
Fig. 6. Slit sector abundances have been divided according to time of
day on Mercury to search for any significant variation in abundance.
There does appear to be a trend for lower abundances toward the early
a.m. sector.
traces along the slit are strongly enhanced at the
illuminated limb and the largest column abundances
are also observed close to the illuminated limb.
Acknowledgements
We wish to thank the TNG personnel for the
competent assistance during these difficult observations.
This work has been partially supported on the Italian
side with a contract from the Ministry of Research and
University (MIUR). At Boston University, this work
was supported by seed research funds from its Center
for Space Physics.
ARTICLE IN PRESS
C. Barbieri et al. / Planetary and Space Science 52 (2004) 1169–1175
Potter, A.E., Morgan, T.H., 1990. Evidence for magnetospheric effects
on the sodium atmosphere of Mercury. Science 248, 835–838.
Potter, A.E., Killen, R.M., Morgan, T.H., 1999. Rapid changes in the
sodium exosphere of Mercury. Planet. Space Sci. 47, 1441–1448.
Potter, A.E., Killen, R.M., Morgan T.H., 2002. The sodium tail of
Mercury. Meteoritics Planetary Sci. 37, 1165–1172.
Smyth, W.H., Marconi, M.L., 1995. Theoretical overview and
modelling of the sodium and potassium atmospheres of Mercury.
Astrophys. J. 441, 839–864.
1175
Sprague, A.L., 1992. Mercury’s atmospheric sodium bright spots: A
possible cause. JGR 97 (Ell), 257–264.
Sprague, A.L., Kozlowski, R.W.H., Hunten, D.M., Schneider, N.M.,
Domingue, D.L., Wells, W.K., Schmitt, W., Fink, U., 1997.
Distribution and abundance of sodium in mercury’s atmosphere,
1985–1988. Icarus 129, 506–527.
Sprague, A.L., Schmitt, W.J., Hill, R.E., 1998. Mercury: sodium
atmospheric enhancements, radar-bright spots and visible surface
features. Icarus 136, 60–68.